Significant Zr Isotope Variations in Single Zircon Grains Recording Magma Evolution History

Significant Zr isotope variations in single zircon grains recording magma evolution history

Jing-Liang Guoa,b,1, Zaicong Wanga,1, Wen Zhanga, Frédéric Moyniera,c, Dandan Cuia, Zhaochu Hua, and Mihai N. Duceab,d

aState Key Laboratory of Geological Processes and Mineral Resources, School of Earth Sciences, China University of Geosciences, Wuhan 430074, China; bDepartment of Geosciences, University of Arizona, Tucson, AZ 85718; cInstitut de Physique du Globe de Paris, Université de Paris, CNRS, 75238 Paris Cedex 05, France; and dFaculty of Geology and Geophysics, University of Bucharest, 010041 Bucharest, Romania

Edited by Bruce Watson, Rensselaer Polytechnic Institute, Troy, NY, and approved July 20, 2020 (received for review February 27, 2020) Zircons widely occur in magmatic rocks and often display internal differentiation processes of the silicate Earth (17, 18), the Zr zonation finely recording the magmatic history. Here, we presented isotopes of zircon might also record related messages. in situ high-precision (2SD <0.15‰ for δ94Zr) and high–spatial-resolution Recent developments of bulk and in situ analytical methods (20 μm) stable Zr isotope compositions of magmatic zircons in a have permitted studies on mass-dependent Zr isotope fraction- suite of calc-alkaline plutonic rocks from the juvenile part of the ation in terrestrial samples, including zircons (15, 19–24). They Gangdese arc, southern Tibet. These zircon grains are internally show the great potential of Zr isotope in tracing magmatic zoned with Zr isotopically light cores and increasingly heavier processes. However, available data have been rather limited and rims. Our data suggest the preferential incorporation of lighter resulted in very controversial interpretation (21, 22), such as for Zr isotopes in zircon from the melt, which would drive the residual fractionation factor α (whether greater or smaller than 1). The α 94 90 94 90 melt to heavier values. The Rayleigh distillation model can well is defined as ( Zr/ Zr)zircon/( Zr/ Zr)melt, which depicts how explain the observed internal zoning in single zircon grains, and Zr isotopes partition into the crystallizing zircon and surround- the best-fit models gave average zircon–melt fractionation factors ing melt and is thus fundamental to understanding magmatic Zr for each sample ranging from 0.99955 to 0.99988. The average isotope evolution. Inglis et al. (21) studied a highly differentiated fractionation factors are positively correlated with the median magmatic suite from the Hekla volcano in Iceland, and suggested Ti-in-zircon temperatures, indicating a strong temperature depen- an α of 0.9995, given the increasing δ94Zr (the permil deviation 94 90 dence of Zr isotopic fractionation. The results demonstrate that of Zr/ Zr from the IPGP-Zr standard) along with decreasing EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES in situ Zr isotope analyses would be another powerful contribu- Zr contents in highly evolved samples (SiO2 >65 wt%). It implies tion to the geochemical toolbox related to zircon. The findings of that zircon favors light Zr isotopes from the melt, consistent with this study solve the fundamental issue on how zircon fractionates theoretical predictions based on the Zr-coordination difference Zr isotopes in calc-alkaline magmas, the major type of magmas betweenzirconandthemelt(21).Ontheotherhand,zirconcrystals that led to forming continental crust over time. The results also extracted from FC1 gabbroic anorthosite (Duluth Complex, north- show the great potential of stable Zr isotopes in tracing magmatic eastern Minnesota) show extremely large Zr isotope variations thermal and chemical evolution and thus possibly continental (∼5‰), revealed by both in situ (15) and bulk measurements (22). crustal differentiation. The majority of bulk-zircon values are heavier than the bulk rock,

Zr isotopes | zircon | magma evolution | laser ablation MC-ICP-MS | Significance crustal differentiation

Zircon, a common accessory mineral in crustal rocks, records ircon (ZrSiO4) is one of the most important accessory min- plentiful and critical information on the Earth’s history. The Zerals in rocks. It occurs in a large variety of lithologies and is isotopes of its major component, Zr, could be another powerful generated primarily in intermediate to felsic igneous rocks (1, 2). but unexplored tracer. We apply high-precision, high–spatial- Its trace element (e.g., Ti and rare-earth element [REE]) and resolution, in situ laser ablation Zr isotope measurements of isotopic (e.g., U–Pb, Lu–Hf, Si, and O) abundances have provided magmatic zircons in continental arc plutonic rocks. Single zir- major information on the absolute timing, petrochronologic con- con grains show impressive internal zoning with lighter Zr ditions, and tectonic backgrounds of important geological events isotopes in the core but heavier ones toward the rim, solving a along the Earth’shistory(3–8). They are especially important for fundamental but controversial issue on how zircon fractionates deciphering the growth and reworking history of the continental Zr isotope in evolving magmas. Our results also reveal a strong crust (9), the early differentiation of the Earth (4, 10), and the temperature dependence of Zr isotopic fractionation. The Zr styles of plate tectonics through time (11, 12). isotope is thus very promising in deciphering the differentia- Magmatic zircons crystallize early in evolved calc-alkaline magmas tion history of magmatic systems and possibly the continental (1). They are relatively insoluble in crustal melts and resistant to crust through time. chemical and physical breakdown. The diffusion of major and Author contributions: J.-L.G., Z.W., and Z.H. designed research; J.-L.G., Z.W., and F.M. trace elements in zircon is also sluggish (13). These characteristics performed research; J.-L.G., W.Z., and D.C. analyzed data; J.-L.G., Z.W., W.Z., F.M., and allow zircon to preserve elemental and isotopic zoning, as well as M.N.D. performed data interpretation; J.-L.G., Z.W., W.Z., F.M., and M.N.D. wrote several generations of geochemical information within a single the paper. grain (2, 13). Compared to bulk analyses, in situ analytical meth- The authors declare no competing interest. ods could extract more detailed thermal and chemical information This article is a PNAS Direct Submission. from intragrain chemical zoning (14, 15). Since zircon has a sig- This open access article is distributed under Creative Commons Attribution License 4.0 (CC BY). nificantly higher Zr concentration (∼48 wt% as a major element) 1To whom correspondence may be addressed. Email: [email protected] or zaicongwang@ than rock-forming minerals (none to a few hundred parts per cug.edu.cn. million) (16), the Zr content of magma should be primarily con- This article contains supporting information online at https://www.pnas.org/lookup/suppl/ trolled by zircon after its saturation. Given that Zr, as one of the high doi:10.1073/pnas.2002053117/-/DCSupplemental. field strength elements, has long been used to monitor magmatic First published August 18, 2020.

www.pnas.org/cgi/doi/10.1073/pnas.2002053117 PNAS | September 1, 2020 | vol. 117 | no. 35 | 21125–21131 Downloaded by guest on September 26, 2021 and an α of 1.00106 was suggested accordingly (22). While these from ∼96 to ∼90 Ma (SI Appendix, Fig. S1 and Table S2), which studies reached opposite conclusions, they both suggested that are consistent with the magmatic flare-up (100 to 85 Ma) of re- zircon crystallization would fractionate the Zr isotope composition gional arc magmatism (26). of the melt. This effect might be recorded in the growth zoning of The Zr isotope compositions of magmatic zircons were ana- single zircon grains. lyzed by femtosecond LA-MC-ICP-MS using a small spot size of The internal growth zoning in magmatic zircons is common (2), 20 μm in diameter, which assured high spatial resolution and high and once formed, it could preserve information on the subtle and precision (SI Appendix,TableS4). Isotope ratios were calibrated transient magmatic differentiation due to the negligible diffusion of against zircon standard GJ-1 or Penglai, which are homogeneous rare earth and high-field strength elements (25). Thus, high–spatial- in Zr isotope compositions shown by Zhang et al. (15). GJ-1 is a resolution in situ analyses of internally zoned zircon grains may gem-quality megacryst zircon from Sri Lanka that has been widely provide key insights into magma differentiation. Here, we apply used for U–Pb and Hf isotope analysis (29). The Penglai zircon is high-precision laser ablation multicollector inductively coupled megacrysts (up to several millimeters in length) from Quaternary plasma mass spectrometry (LA-MC-ICP-MS) and report in situ Zr potassic basalts in south China used for O and Hf isotope analysis isotope compositions of magmatic zircons from mafic to inter- (30). In order to directly compare our zircon data with previous mediate plutonic rocks, which represent the juvenile part of the bulk analyses, the 94Zr/90Zr ratios were further expressed as the Gangdese arc in southern Tibet (26, 27). The results reveal large permil deviation from solution standard IPGP-Zr (Plasma-Cal internal Zr isotope variations in single grains, showing unambiguously standard solution; lot 5131203028). This standard has been used the preference of light Zr isotope by zircons and set the base for wide as a reference standard in most studies reporting Zr isotopic com- future applications. This study further indicates a strong temperature positions and was recently cross-calibrated with different geological – influence on zircon melt Zr isotope fractionation, suggesting its po- reference materials (15, 20, 21, 23). The within-run and long-term tential in depicting magmatic thermal and chemical evolution. external precision for δ94Zr in this study are better than 0.10‰ ‰ δ94 Mafic–Intermediate Plutonic Rocks and Zircons from the (2SE) and 0.15 (2SD), respectively. The Zr values of all an- − ‰ ‰ Gangdese Arc, Southern Tibet alyzed zircons range from 0.86 to 0.41 ,withanaverage of −0.04 ± 0.39‰ (2SD, n = 237). The overall range (1.27‰)is The Gangdese arc in southern Tibet was produced by the north- larger than recently reported bulk rocks (0.582‰)(21),butsmaller ward subduction of the Neo-Tethyan oceanic lithosphere during than that (∼5‰) observed in zircon and baddeleyite crystals from the Mesozoic; this arc later witnessed the India–Asia collision during the Cenozoic (26, 28). Large volumes of Mesozoic plutonic anorthosite FC1 (15, 22). Most zircon grains in this study show well-developed internal rocks outcropped along the eastern segment of the Gangdese arc. isotope zoning (Figs. 2 and 3). The intragrain variations can be as These rocks typically show depleted mantle-like isotopic features, ‰ ± ‰ n = indicating significant juvenile crustal growth during this period significant as 0.74 , with an average of 0.31 0.38 (2SD, – 38) (SI Appendix, Table S4). The isotope profiles are characterized (26). The studied samples belong to a suite of mafic intermediate δ94 plutonic rocks collected from the Quxu batholith in the eastern by low Zr in the core and elevated values toward the rim (Figs. δ94 Gangdese arc, including hornblende gabbros, tonalites, and 2 and 3). The average Zr value of zircon core for each sample – biotite-rich microgranular enclaves hosted by the tonalites (Fig. 1 shows a negative correlation with the relative core rim difference Δ SI Appendix and SI Appendix,TableS1). These samples represent the juvenile ( Rim–Core)( ,TableS1). A few zircon grains show δ94 part of Gangdese arc crust formed mainly by the fractional crys- complex zoning patterns with elevated or reduced Zr values at G J L tallization of arc magmas (27). Zirconium isotopes of the zircons the mantle relative to the core (e.g., Fig. 2 , ,and ). Some of L may have recorded the differentiation of those calc-alkaline arc them show truncation surfaces in the core (e.g., Fig. 2 ), indicating plutonic rocks. Here, we analyzed their U–Pb ages, major and a possible antecrystic origin (inherited from previous magmatic trace element compositions, as well as Zr isotope compositions. pulses) given the indistinguishable ages (31). Zircons are rare in hornblende gabbros but are common in Major and trace element profiles were also obtained along the tonalites and biotite-rich enclaves. Most extracted grains are Zr isotope profiles (SI Appendix, Table S5). The Zr/Hf ratios 94 ∼100 to 300 μm in length with euhedral shapes. They vary in size generally decrease from core to rim, opposite to δ Zr (Fig. 2). and internal structure among different rock types (Fig. 2). Zir- This trend is clearer for zircons in samples with lower magmatic 94 cons from the hornblende gabbros are generally smaller and have temperatures, and the slopes of δ Zr vs. Zr/Hf for single zircon linear, sector, or wide-banded oscillatory zoning (Fig. 2 A–F). grains become steeper from hornblende gabbros to tonalites and In contrast, zircons from the tonalites and the biotite-rich the biotite-rich enclave (SI Appendix, Fig. S5). The Zr/Hf ratio of enclave are larger and show dominantly oscillatory zoning with zircon has been used as an indicator for magmatic differentiation; thin bands (Fig. 2 G–J). These features, together with high Th/U zircons from more evolved melts tend to have lower Zr/Hf ratios ratios (0.4 to 2.9) (SI Appendix, Table S2), indicate a magmatic (32). The broadly negative correlation between δ94Zr values and origin (2). No systematic U–Pb age difference was found between Zr/Hf ratios implies that the Zr isotope variations in zircon also the core and the rim. The emplacement ages of all samples range records magmatic differentiation related to zircon crystallization.

Hb gabbro Tonalite Bi-rich enclave A B C DEF 1 c m

D17T107D17T100 D17T102 D17T091 D17T119 D17T109

Fig. 1. Scanned images of thick petrographic sections for plutonic rocks from the Gangdese arc, southern Tibet. (A–C) Hornblende gabbros. (D and E) Tonalites. (F) Biotite-rich enclave. Bi, biotite; Hb, hornblende.

21126 | www.pnas.org/cgi/doi/10.1073/pnas.2002053117 Guo et al. Downloaded by guest on September 26, 2021 AB C

DE F

GH I

JKL

Fig. 2. Representative Zr isotope and Zr/Hf profiles of magmatic zircon in plutonic rocks from the Gangdese arc, southern Tibet. (A–F) Hornblende gabbros. EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES (G–J) Tonalites. (K and L) Biotite-rich enclave. The solid black and dashed red lines indicate δ94Zr and Zr/Hf profiles, respectively. The solid blue curves denote best-fit isotope profiles modeled from the Rayleigh distillation equation. The dotted gray lines indicate the initial melt composition used for modeling. Cathodoluminescence images of zircon are presented on the Left, showing the internal structures. The solid yellow, dotted red, and dashed white circles denote LA spots for Zr isotopes (20 μm in diameter), Zr/Hf profiles (24 μm in diameter), and U–Pb isotopes and trace elements (32 μm in diameter), respectively. Error bars for the δ94Zr and Zr/Hf ratios represent ±2SE uncertainties.

The crystallization temperature of zircon was estimated using has quartz and rutile (inclusions in biotite), and thus the activi- the Ti-in-zircon thermometer (33) (SI Appendix, Tables S2 and ties of SiO2 and TiO2 were both assigned to 1. The obtained S5). This thermometer requires inputs of zircon Ti concentra- temperatures were further corrected to 0.4 GPa (corresponding tions, bulk-rock SiO2 and TiO2 activities, and the pressure. The to a temperature increase of 30 °C) (33). The studied samples activities have a critical impact on the estimated temperatures show systematic differences in the median Ti-in-zircon temper- (34), which were calculated using the thermodynamic software ature: 743 to 831 °C for hornblende gabbros, 726 to 742 °C for Perple_X (35) except for the biotite-rich enclave. This sample tonalites, and 689 °C for the biotite-rich enclave (SI Appendix,

BCD E F G

Fig. 3. Intragrain core-to-rim Zr isotope variations in zircon from plutonic rocks in the Gangdese arc, southern Tibet. (A) Three representative zircon grains for each sample. (B–G) Histograms of δ94Zr values for the core, mantle, and rim of all zircon grains in each sample. N is the number of total analyses for each sample. The dotted red lines denote the primitive mantle value of 0.048 ± 0.032‰ (2SD) (21). Error bars represent ±2SE uncertainty.

Guo et al. PNAS | September 1, 2020 | vol. 117 | no. 35 | 21127 Downloaded by guest on September 26, 2021 Table S1). The temperature differences in different samples have influenced the Zr isotope composition of FC1 zircons and the are consistent with their lithologies and petrographic textures bulk rock. More work is needed to answer this question. (SI Appendix). The internal zoning in zircon can provide more direct evidence and avoid some of the potential problems faced by bulk analyses Significance of Internal Zr Isotope Zoning in Single Zircon (15, 24). The observed isotope profiles can be further modeled Grains by Rayleigh distillation (Fig. 2 and SI Appendix, Text S6). This The most significant finding in this study is the large internal Zr model assumes 1) equilibrium at the zircon–melt interface and 2) isotope zoning (up to 0.74‰) in single zircon grains from the the melt is homogeneous, but once zircon is crystallized, it is studied arc plutonic rocks (Figs. 2 and 3). Chemical zoning in considered to be isolated from the system. Assuming zircon as minerals could be induced by growth or subsequent diffusion. the only mineral controlling Zr isotope fractionation, the com- The diffusivity of Zr in zircon was inferred from its geochemical position of crystallizing zircon and the residual melt (Fig. 4A) can twin Hf, which is almost immobile in zircon even at temperatures be modeled using the following equations: of 900 to 1,000 °C (25). Therefore, the Zr isotope zoning in zircon reflects most likely growth zoning, which may record the change of 94 94 ()α−1 δ Zrzircon = ()δ Zrmelt,0 + 1, 000 α F − 1, 000, [1] melt composition during magma differentiation.

Isotope profiles in these zircon grains are characterized by light ()α− 94 δ94 = ()δ94 + 1 − [2] cores and gradually heavier rims (Fig. 2). Most δ Zr values of the Zrmelt Zrmelt,0 1, 000 F 1, 000, zircon cores (as well as the average value of all analyses) are lower δ94 α than that of the primitive mantle (0.048 ± 0.032‰) (21). Such where Zrmelt,0 is the initial melt isotopic composition, is the zoning patterns indicate the preferential incorporation of light Zr zircon–melt isotopic fractionation factor, and F stands for the isotopes into zircon (i.e., α < 1). Our finding is consistent with the fraction of Zr content remaining in the melt. If α < 1, the crys- first-order theoretical prediction where higher coordination-number tallizing zircon is isotopically lighter than the surrounding melt; lattices will be enriched in lighter isotopes, since in zircon, Zr has while zircon grows, the melt would evolve to heavier values, and eightfold coordination (36), while in the melt it is generally found so would the lately formed zircon rims (Fig. 4A). The Zr isotope withsixfoldcoordination(37,38).Thisisalsoconsistentwiththeα profile in zircon can be further modeled by assuming sphere of 0.9995 inferred from bulk analyses of the Hekla volcanic rocks, shapes for both zircons and the melt cell (24, 41). The radius − where rhyolites (SiO >65 wt%) show elevated δ94Zr values with of zircon grain would be proportional to the cubic root of 1 F 2 SI Ap- reduced Zr contents (21). It is, however, different from the α of (i.e., the fraction of Zr consumed by zircon) (Fig. 2 and pendix 1.00106 suggested from FC1 zircons in the Duluth anorthosite (15, ). More complex situations can be modeled by using a 22). The FC1 zircons exhibit a much larger δ94Zr range of ∼5‰ recently published MATLAB routine dealing with zircon disso- (15,22)thanobservedinmegacrystzircons(<0.3‰) (15, 24) or lution and crystallization in complex magmatic systems (41). δ94 fine- to medium-grained zircons in this study (1.27‰). The ma- The fractionation factor controls not only the Zr difference jority of FC1 zircons are isotopically heavier than the bulk rock by between the crystallizing zircon and the surrounding melt, but also α ∼0.3‰,resultinginα > 1 (22). However, the large range in FC1 the shape of isotope profile (the closer the is to 1, the flatter the was mainly caused by two low values (−2.530‰ and −4.278‰)in profile is) (Fig. 2). Both factors were considered to determine the zircon fragments that were not treated by chemical abrasion (a best-fit model, from which α can be estimated (Fig. 2). In our method to selectively remove domains with high accumulated modeling, we avoided zircon cores that are inherited in origin. δ94 = ‰ ‰ radiation damage) (22). If they were excluded, the range would Different initial melt compositions ( Zrmelt, 0 0.05 ,0.15 , drop to ∼1.8‰ (vs. 1.27‰ in this study). One anomalously low and 0.25‰) were tested. The value of 0.05‰ was preferred value of −3.44‰ also exists in in situ data, in contrast to the rest unless higher values provided better-fitting curves for the observed ranging from −0.84‰ to 1.49‰ (15). Radiation damage and isotope profiles, especially for those with overall heavy isotope later hydrothermal alteration were actually documented in anor- compositions (Fig. 2 and SI Appendix, Table S6). thosite zircon AS3 from another outcrop in the Duluth Complex, The assumption of a homogeneous melt for Zr in the Rayleigh resulting in depleted ZrO2 and SiO2 contents, discordant U–Pb distillation model may seem to be problematic, because its dif- ages, and enriched nonformula elements in metamict domains fusivity (D) of Zr is low in the melt (42, 43) and kinetic isotope (39). Slight age discordance has also been reported for FC1 zircons fractionation may take place in the diffusive boundary melt layer (40). It is unclear whether metamictization (and later hydrothermal near zircon (44). Since light isotopes diffuse faster, they may lead alteration) or the early crystallization of other Zr-bearing minerals to the observed isotopically light cores of zircon. However, the

Fig. 4. The fractionation mechanism of Zr isotopes between zircon and melt. (A) Rayleigh distillation modeling of Zr isotope evolution of crystallizing zircons and the residual melts. The solid and dashed lines denote zircon and melt compositions, respectively. The initial melt composition was assumed to be 0.05‰. (B) Correlation between average fractionation factor α and median Ti-in-zircon temperatures (T) for the studied samples. Fractionation factors were estimated from the best-fit Rayleigh distillation models. The overall trend indicates a strong temperature dependence, but large uncertainty exists in the regression, and cautions are needed when applying the obtained empiric equation. Error bars represent ±2SE uncertainty. The shaded area indicates the 95% confidence interval.

21128 | www.pnas.org/cgi/doi/10.1073/pnas.2002053117 Guo et al. Downloaded by guest on September 26, 2021 growth rate (R) of zircon is normally low (45, 46), which may correlation (square of the correlation coefficient, r2 = 0.86) be- − reasonably result in R/D ratios <0.01 cm 1 and significant kinetic tween 1,000 ln(α) and 1/T2 (T in kelvin) is obtained as follows: isotope fractionation is not expected unless R/D is much higher (44). Nevertheless, zircon growth rates and Zr diffusivities in the 1, 000 ln()α = –0. 95 ± 0. 84 × 106=T2 + 0. 65 ± 0. 81. [4] melt are not constant. If kinetic isotope effect was in control, the produced isotope profile in single zircon grain would possibly The correlation relies on the estimated fractionation factor, and have a “W” shape as shown for trace elements based on a diffusion- thus the shape of Zr isotope profiles and the assumed initial melt controlled crystal growth model (44), being characterized by an compositions. The nonzero free term B is mostly caused by isotopically heavy core, a light mantle, and a heavy rim (given a uncertainties in our data (the sample size is not large enough at finite melt cell). The estimated α for zircons with W-shaped isotope present) and in the formula extrapolation. Less likely, it may be profiles could be underestimated. However, zircon δ94Zr profiles in induced by zircon formation at nonequilibrium conditions for this study mostly exhibit simple U shapes, and only a few have W samples with low magmatic temperatures. Even if the latter were shapes (Fig. 2 G and J). This indicates that if kinetic isotope frac- the case, the overall condition should be still close to equilibrium, tionation had existed, its role was limited. The observed isotope as indicated by the U-shaped isotope profiles in most zircon grains. Given the large uncertainty in the regression, this empiric profiles are still best portraited by the Rayleigh distillation model as Eq. 4 so far cannot be applied to estimate precise temperature, showninFig.2. which requires future calibration. Nonetheless, the good correla- The average α values for all six samples range from 0.99955 to tion (r2 = 0.86) itself still indicates a strong temperature depen- 0.99988 (Fig. 2 and SI Appendix,TableS6), closer to 0.9995 inferred dence of zircon–melt Zr isotope fractionation. from Hekla volcanic rocks (21). The obtained values are all below 1, The empirical Eq. 4 indicates that α will drop from 0.99983 to confirming the preference of light Zr isotopes in zircon. Moreover, α 0.99965 when the temperature decreases from 800 to 700 °C. At varies systematically among the hornblende gabbros (0.99974 to temperatures >850 °C, limited isotope fractionation (<0.1‰)is 0.99988), tonalites (0.99974 to 0.99978), and biotite-rich enclave expected, whereas at lower temperatures, fractionation is supposed B (0.99955) (Fig. 4 ). This suggests that the Zr isotope evolution of to be more significant especially for felsic magmas that may reach α magmatic systems is a dynamic process, and a constant may not early zircon saturation. This equation also implies the potential use be able to fully describe this process. of simple zircon Zr isotope zoning as a tracer for changes in magma temperatures. Another application would be the estimation of ini- Temperature Dependence of Zr Isotope Fractionation 94 94 tial melt composition [δ Zr = δ Zr − 1,000 ln(α), – α melt, 0 zircon, core As shown above, the zircon melt fractionation factor may vary and it is also noted that the rim composition is generally close to the EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES in magmas of different temperatures and compositions. It is assumed initial melt in Fig. 2] if the temperature can be constrained important to assess the potential controlling factor(s) on α.As by other thermometers (Fig. 4B), which may be applicable to trace predicted by classic stable isotope fractionation theories (47), mass- magmatic Zr isotope evolution. dependent fractionation should have a strong temperature dependence. At high-temperature conditions, the relationship between α Tracing Magmatic Evolution History and Continental Crustal for thermodynamic fractionation and the temperature (T, in kelvin) Differentiation is described as follows: Our study shows the importance of in situ zircon Zr isotope analyses in understanding the chemical and thermal evolution of 1, 000 ln(α) = A × 106=T2 + B, [3] calc-alkaline magmatic systems. These systems have played a major role in producing the continental crust (48). Magmatic zircon can where A and B are equilibrium constants, and B should be 0 since preserve growth zoning of both Zr isotopes (Fig. 2) and trace ele- fractionation approaches 0‰ at infinitely high temperatures. ments (49) due to its low solubility in crustal melts, high resistance We use the average α value and median Ti-in-zircon temperature to chemical/physical breakdown, and low diffusivities of trivalent of zircon to evaluate the temperature dependence (Fig. 4B). The and tetravalent cations (e.g., REE, Ti, Zr, Hf, U, and Th) (25, 50).

Fig. 5. Schematic of zircon-induced Zr isotope fractionation during magma differentiation and examples of the produced Zr isotope profiles in single zircon grains, showing the potential of Zr isotopes in tracing the differentiation of magmatic systems and possibly the continental crust. The isotope profiles of zircon were modeled from the Rayleigh distillation equation, assuming an initial melt composition of 0.05‰ (dotted gray lines) with fractionation factor α calculated from Eq. 4 at given temperatures. (A) Light Zr isotopes in calc-alkaline magmas are preferentially incorporated by zircon, driving the residual melt to isotopically heavier compositions. Due to the strong temperature (T) dependence of zircon–melt Zr isotope fractionation, larger isotope fractionation is expected when magmas evolve to lower temperatures. (B) Modeled Zr isotope profile in zircon crystallized at 700 °C with a simple growth history. (C) Modeled Zr isotope profile in zircon with a complex growth history. The rim (solid blue line) and the core (dashed orange line) were calculated at temperatures of 700 and 800 °C, respectively. Note that complex zoning patterns may exist in natural zircons. (D) Modeled Zr isotope profile of zircon crystallized at 800 °C with a simple growth history.

Guo et al. PNAS | September 1, 2020 | vol. 117 | no. 35 | 21129 Downloaded by guest on September 26, 2021 Unlike the trace element zoning that is controlled by multiple Conclusions factors (such as the pressure and temperature, melt composition, We present in situ Zr isotope compositions of magmatic zircon in coexisting minerals, and the growth rate of zircon) (49, 51), the Zr a calc-alkaline plutonic suite from the juvenile part of the Gangdese isotope zoning in magmatic zircons is less complex and mainly arc, southern Tibet. The Zr isotope data were obtained by LA- B controlled by the melt composition and temperature (Figs. 4 and MC-ICP-MS with a high spatial resolution (20 μm) and high pre- 5). Technically, it is also much easier to achieve high spatial res- cision (2SD <0.15‰ for δ94Zr). The results show large variations olution and high precision for LA in situ Zr isotope measurements ranging from −0.86‰ to 0.41‰, and most individual zircon grains as Zr is a major element of zircon with limited isotopic interfer- exhibit internal zoning with low δ94Zr in the core and higher values ences (this study and ref. 15). For zircon grains with simple growth history (Fig. 5 B and D), their internal Zr isotope profiles can be toward the rim. This indicates that zircon favors light Zr isotopes used to trace changes in the magma temperature by applying the from the melt, and its crystallization would drive melt to heavier Zr empirically calibrated correlation between zircon–melt α and the isotope compositions. This would set the base for future interpre- magma temperature. For zircons with relatively complex growth tation of the Zr isotope evolution in calc-alkaline igneous rocks, the history (e.g., Fig. 5C), differences between anatectic cores and major constituent of the continental crust. The observed isotope autocrystic rims may reveal additional information on the magmatic profiles in single zircon grains can be well explained by the Ray- evolution history. For example, an anatectic core, with a truncation leigh distillation model. The best-fit models gave different average surface in CL images and a higher δ94Zr value, may have formed at α values (0.99955 to 0.99988) for the studied samples, which are a higher temperature than the rim (Fig. 5C). Together with other positively correlated with the median Ti-in-zircon temperatures, zircon-based geochemical tools, stable Zr isotopes would have wide indicating a strong temperature dependence of zircon–melt Zr applications in depicting magmaticprocessessuchasmagmaas- isotope fractionation. The results also demonstrate that in situ similation and recharge or preeruption volcanic processes. Zr isotope profiles in zircon could provide key insights into the The internal isotope profile of single magmatic zircon further chemical and thermal history of magmatic systems. The stable Zr suggests that zircon preferentially incorporates light Zr isotopes, isotope is thus a promising powerful tracer in revealing the dif- which means that the heavy isotopes would be concentrated in ferentiation of terrestrial magmatic systems and the Earth’s the residual melt. Zirconium, as an incompatible element during continental crust. mantle melting and early magma differentiation (17, 18), is fa- vored by the melt that eventually formed the continental crust of Methods the Earth (48). Large fractionation among its isotopes is sup- The stable Zr isotopes of zircon were analyzed by using an NWRfemto posed to occur at late stages when the residual melts evolve to > – femtosecond LA system (New Wave Research) coupled with a Neptune Plus felsic compositions ( 65 wt% SiO2) by liquid crystal segregation MC-ICP-MS (Thermo Fisher Scientific) at the China University of Geosciences, (18, 21). Since zircon is generally unsaturated in oceanic basalts Wuhan. The spot size was set to 20 μm in diameter. The ablation frequency ∼ ∼ for their mafic (SiO2 50 wt%), low-Zr ( 70 ppm), and high- was as low as 1 Hz, which reduced the risk of isotope fractionation induced temperature nature (52), large Zr isotope fractionation is not by small-beam LA. Zircon U–Pb dating and trace element analyses were expected for the oceanic crust. Even if zircon exists, its crystalli- performed by using a 193-nm excimer ArF LA system (GeoLas HD; Coherent) zation temperature would probably be high and large Zr isotope coupled with Agilent 7700x quadrupole ICP-MS. Major and trace element fractionation is still not expected. This is consistent with the ob- compositions of zircon were further analyzed next to Zr isotope profiles by served narrow variation range of δ94Zr (−0.033‰ to 0.086‰)in using GeoLas Pro 193-nm excimer ArF LA system coupled with Agilent 7700e oceanic basalts (21). In comparison, the continental crust is char- ICP-MS at the Wuhan SampleSolution Analytical Technology Company. The acterized by an average andesitic composition (SiO2 = ∼60 wt%) CL images of zircon were taken to reveal the internal structure of zircon by (48) with more differentiated intermediate to felsic magmatic using Analytical Scanning Electron Microscope JSM-IT100 (InTouchScope; rocks. These rocks are easier to reach early zircon saturation than JEOL) connected to a MINICL system (Gatan). The bulk-rock major and trace mafic ones (1). If the isotopically light zircon can be effectively element compositions were analyzed by using X-ray fluorescence and removed, rocks like highly differentiated granites and rhyolites Agilent 7900 quadrupole ICP-MS, respectively. Analytical details can be formed from the residual melt would have heavier Zr isotope found in SI Appendix. compositions. The upper continental crust, where evolved magmas are concentrated, thus may have a heavier average Zr isotope Data Availability. All study data are included in SI Appendix. composition than the primitive mantle. When the crust becomes ACKNOWLEDGMENTS. We are grateful to two anonymous reviewers for more mature, the enrichment degree of heavy Zr isotopes in the constructive suggestions that greatly improved the manuscript. We thank upper crust would probably also increase. Thus, the Zr isotope Tao Luo for assistance in the LA-ICP-MS laboratory. This study was supported composition of the upper crust in different continental regions may by grants from the National Key Research and Development Project of China be used as an indicator for crustal maturation. Once confirmed, (2016YFC0600309), the National Natural Science Foundation of China (41973024 zircon Zr isotopes may open a new and insightful way to under- and 41730211), the 111 Project (BP0719022), the Fundamental Research Funds for stand the evolution of the continental crust over geological time the Central Universities (CUGL180413), and the most special fund from the State Key Laboratory of Geological Processes and Mineral Resources (GPMR202002). (from 4.4 billion years by Jack Hill zircons to present), particularly M.N.D. acknowledges support from NSF Grant EAR 1725002 and the Romanian at the early stage of the first one or two billion years when the Executive Agency for Higher Education, Research, Development and Innovation geological record but zircon has been relatively rare. Funding Project PN-III-P4-ID-PCCF-2016-0014.

1. E. B. Watson, T. M. Harrison, Zircon saturation revisited: Temperature and composi- 7. D. Trail et al., Origin and significance of Si and O isotope heterogeneities in Phanerozoic, tion effects in a variety of crustal magma types. Earth Planet. Sci. Lett. 64, 295–304 Archean, and Hadean zircon. Proc. Natl. Acad. Sci. U.S.A. 115,10287–10292 (2018). (1983). 8. J. W. Valley, Oxygen isotopes in zircon. Rev. Mineral. Geochem. 53, 343–385 (2003). 2. F. Corfu, J. M. Hanchar, P. W. O. Hoskin, P. Kinny, Atlas of zircon textures. Rev. 9. C. J. Hawkesworth, A. I. S. Kemp, Using hafnium and oxygen isotopes in zircons to Mineral. Geochem. 53, 469–500 (2003). unravel the record of crustal evolution. Chem. Geol. 226, 144–162 (2006). 3. C. Balica et al., A zircon petrochronologic view on granitoids and continental evolu- 10. A. I. S. Kemp et al., Hadean crustal evolution revisited: New constraints from Pb–Hf tion. Earth Planet. Sci. Lett. 531, 116005 (2020). isotope systematics of the Jack Hills zircons. Earth Planet. Sci. Lett. 296,45–56 (2010). 4. T. M. Harrison, The Hadean crust: Evidence from >4 Ga zircons. Annu. Rev. Earth 11. M. Hopkins, T. M. Harrison, C. E. Manning, Low heat flow inferred from >4 Gyr zircons Planet. Sci. 37, 479–505 (2009). suggests Hadean plate boundary interactions. Nature 456, 493–496 (2008). 5. T. M. Harrison, E. A. Bell, P. Boehnke, Hadean zircon petrochronology. Rev. Mineral. 12. N. M. W. Roberts, C. J. Spencer, The zircon archive of continent formation through Geochem. 83, 329–363 (2017). time. Geol. Soc. Spec. Publ. 389, 197–225 (2015). 6. D. Rubatto, Zircon trace element geochemistry: Partitioning with garnet and the link 13. D. J. Cherniak, E. B. Watson, Diffusion in zircon. Rev. Mineral. Geochem. 53, 113–143 between U–Pb ages and metamorphism. Chem. Geol. 184, 123–138 (2002). (2003).

21130 | www.pnas.org/cgi/doi/10.1073/pnas.2002053117 Guo et al. Downloaded by guest on September 26, 2021 14. T. M. Harrison, A. K. Schmitt, High sensitivity mapping of Ti distributions in Hadean 32. L. L. Claiborne et al., Tracking magmatic processes through Zr/Hf ratios in rocks and Hf zircons. Earth Planet. Sci. Lett. 261,9–19 (2007). and Ti zoning in zircons: An example from the Spirit Mountain batholith, Nevada. 15. W. Zhang et al., Determination of Zr isotopic ratios in zircons using laser-ablation Mineral. Mag. 70, 517–543 (2006). multiple-collector inductively coupled-plasma mass-spectrometry. J. Anal. At. Spec- 33. J. M. Ferry, E. B. Watson, New thermodynamic models and revised calibrations for the trom. 34, 1800–1809 (2019). Ti-in-zircon and Zr-in-rutile thermometers. Contrib. Mineral. Petrol. 154, 429–437 16. F. Bea, P. Montero, M. Ortega, A LA–ICP–MS evaluation of Zr reservoirs in common (2007). crustal rocks: Implications for Zr and Hf geochemistry, and zircon-forming processes. 34. D. Schiller, F. Finger, Application of Ti-in-zircon thermometry to granite studies: Can. Mineral. 44, 693–714 (2006). Problems and possible solutions. Contrib. Mineral. Petrol. 174, 51 (2019). 17. A. W. Hofmann, Chemical differentiation of the Earth: The relationship between 35. J. A. D. Connolly, K. Petrini, An automated strategy for calculation of phase diagram sections and retrieval of rock properties as a function of physical conditions. mantle, continental crust, and oceanic crust. Earth Planet. Sci. Lett. 90, 297–314 J. Metamorph. Geol. 20, 697–708 (2002). (1988). 36. F. Farges, The structure of metamict zircon: A temperature-dependent EXAFS study. 18. C.-T. A. Lee, O. Bachmann, How important is the role of crystal fractionation in Phys. Chem. Miner. 20, 504–514 (1994). making intermediate magmas? Insights from Zr and P systematics. Earth Planet. Sci. 37. F. Farges, C. W. Ponader, G. E. Brown, Structural environments of incompatible ele- Lett. 393, 266–274 (2014). ments in silicate glass/melt systems: I. Zirconium at trace levels. Geochim. Cosmochim. 19. W. Akram, M. Schönbächler, Zirconium isotope constraints on the composition of Acta 55, 1563–1574 (1991). – Theia and current Moon-forming theories. Earth Planet. Sci. Lett. 449, 302 310 (2016). 38. T. Tobase et al., XAFS study on the Zr local structures in tektites and natural glasses. 20. E. C. Inglis, J. B. Creech, Z. Deng, F. Moynier, High-precision zirconium stable isotope J. Mineral. Petrol. Sci. 110,1–7 (2015). measurements of geological reference materials as measured by double-spike 39. M. Takehara, K. Horie, T. Hokada, S. Kiyokawa, New insight into disturbance of U-Pb – MC-ICPMS. Chem. Geol. 493, 544 552 (2018). and trace-element systems in hydrothermally altered zircon via SHRIMP analyses of 21. E. C. Inglis et al., Isotopic fractionation of zirconium during magmatic differentiation zircon from the Duluth Gabbro. Chem. Geol. 484, 168–178 (2018). and the stable isotope composition of the silicate Earth. Geochim. Cosmochim. Acta 40. D. Frei, A. Gerdes, Precise and accurate in situ U–Pb dating of zircon with high sample 250, 311–323 (2019). throughput by automated LA-SF-ICP-MS. Chem. Geol. 261, 261–270 (2009). 22. M. Ibañez-Mejia, F. L. H. Tissot, Extreme Zr stable isotope fractionation during 41. I. N. Bindeman, O. E. Melnik, Zircon survival, rebirth and recycling during crustal magmatic fractional crystallization. Sci. Adv. 5, eaax8648 (2019). melting, magma crystallization, and mixing based on numerical modelling. J. Petrol. 23. L. Feng et al., High-precision stable zirconium isotope ratio measurements by double 57, 437–460 (2016). spike thermal ionization mass spectrometry. J. Anal. At. Spectrom. 35, 736–745 (2020). 42. T. M. Harrison, E. B. Watson, Kinetics of zircon dissolution and zirconium diffusion in 24. H. G. D. Tompkins, L. J. Zieman, M. Ibañez-Mejia, F. L. H. Tissot, Zirconium stable granitic melts of variable water content. Contrib. Mineral. Petrol. 84,66–72 (1983). isotope analysis of zircon by MC-ICP-MS: Methods and application to evaluating intra- 43. M. E. Holycross, E. B. Watson, Trace element diffusion and kinetic fractionation in wet crystalline zonation in a zircon megacryst. J. Anal. At. Spectrom. 35, 1167–1186 (2020). rhyolitic melt. Geochim. Cosmochim. Acta 232,14–29 (2018). 25. D. J. Cherniak, J. M. Hanchar, E. B. Watson, Diffusion of tetravalent cations in zircon. 44. E. B. Watson, T. Müller, Non-equilibrium isotopic and elemental fractionation during Contrib. Mineral. Petrol. 127, 383–390 (1997). diffusion-controlled crystal growth under static and dynamic conditions. Chem. Geol. – 26. W.-Q. Ji, F.-Y. Wu, S.-L. Chung, J.-X. Li, C.-Z. Liu, Zircon U–Pb geochronology and Hf 267, 111 124 (2009). isotopic constraints on petrogenesis of the Gangdese batholith, southern Tibet. 45. E. B. Watson, Surface enrichment and trace-element uptake during crystal growth. – Chem. Geol. 262 , 229–245 (2009). Geochim. Cosmochim. Acta 60, 5013 5020 (1996). 46. A. K. Schmitt et al., Acigöl rhyolite field, Central Anatolia (part 1): High-resolution 27. W.-C. Xu, H.-F. Zhang, B.-J. Luo, L. Guo, H. Yang, Adakite-like geochemical signature EARTH, ATMOSPHERIC, AND PLANETARY SCIENCES dating of eruption episodes and zircon growth rates. Contrib. Mineral. Petrol. 162, produced by amphibole-dominated fractionation of arc magmas: An example from 1215–1231 (2011). the Late Cretaceous magmatism in Gangdese belt, south Tibet. Lithos 232, 197–210 47. H. C. Urey, The thermodynamic properties of isotopic substances. J. Chem. Soc., (2015). 562–581 (1947). 28. A. Yin, T. M. Harrison, Geologic evolution of the Himalayan-Tibetan orogen. Annu. 48. R. L. Rudnick, S. Gao, “4.1—composition of the continental crust” in Treatise on Rev. Earth Planet. Sci. 28, 211–280 (2000). Geochemistry, H. D. Holland, K. K. Turekian, Eds. (Elsevier, ed. 2, 2014), pp. 1–51. 29. S. E. Jackson, N. J. Pearson, W. L. Griffin, E. A. Belousova, The application of laser 49. P. W. O. Hoskin, Patterns of chaos: Fractal statistics and the oscillatory chemistry of – ablation-inductively coupled plasma-mass spectrometry to in situ U Pb zircon geo- zircon. Geochim. Cosmochim. Acta 64, 1905–1923 (2000). – chronology. Chem. Geol. 211,47 69 (2004). 50. D. J. Cherniak, Si diffusion in zircon. Phys. Chem. Miner. 35, 179–187 (2008). 30. X. H. Li et al., Penglai zircon megacrysts: A potential new working reference material 51. O. E. Melnik, I. N. Bindeman, Modeling of trace elemental zoning patterns in acces- for microbeam determination of Hf–O isotopes and U–Pb age. Geostand. Geoanal. sory minerals with emphasis on the origin of micrometer-scale oscillatory zoning in Res. 34, 117–134 (2010). zircon. Am. Mineral. 103, 355–368 (2018). 31. C. Siégel, S. E. Bryan, C. M. Allen, D. A. Gust, Use and abuse of zircon-based ther- 52. S. S. Sun, W. F. McDonough, Chemical and isotopic systematics of oceanic basalts: mometers: A critical review and a recommended approach to identify antecrystic Implications for mantle composition and processes. Geol. Soc. Spec. Publ. 42, zircons. Earth Sci. Rev. 176,87–116 (2018). 313–345 (1989).

Guo et al. PNAS | September 1, 2020 | vol. 117 | no. 35 | 21131 Downloaded by guest on September 26, 2021